AI-assisted development with validation loops
How to actually use AI agents in a real codebase without handing over the steering wheel. We cover the validation loops, review habits, and small-diff discipline that keep generated work honest. An AI development workshop framing, grounded in code you'd be willing to ship.
Best for
Engineering teams adopting AI tooling who want speed without losing trust in the output.
AI-Assisted Development Without Losing Architecture → Why AI acceleration is not architecture
AI makes typing faster; it does not make decisions for you. This talk separates raw output speed from the design judgment that decides whether a system holds together, and where teams confuse the two. The fix is narrow, validated slices instead of heroic big-bang generation.
Best for
Founders and leads deciding how much of their build to hand to AI.
Small Slices Beat Big Bang AI → Safe automation: dry runs, logs, rollback, and human approval
Automation without a panic button is just a faster mistake. We walk through dry-run mode, logs that mean something, least-privilege scope, rollback thinking, and explicit confirmation on anything destructive. A safe automation conversation you can apply the same week.
Best for
Ops-minded builders and teams automating real, consequential systems.
Automation Needs a Panic Button → Turning messy ideas into small buildable slices
Most good builds start as something half-formed and a little too big. This session is about shrinking a vague idea to the narrowest slice that does something real, then iterating from there. It's the difference between a roadmap you trust on faith and a step you can take today.
Best for
Anyone sitting on an idea that feels too big to start.
Building Weird Ideas Into Real Systems → Software architecture that survives implementation
A clean diagram is easy; an architecture that's still clean after three months of real work is the hard part. We talk about boundaries, naming, and the judgment calls that decide whether a system stays maintainable. A software architecture talk aimed at decisions, not diagrams.
Best for
Software and product teams who want structure that holds up under change.
Security as an architecture input
Security works best as a design discipline, not a bolt-on at the end. This is a cybersecurity discussion about threat modeling, least privilege, and the boring-but-right habits that belong in the design phase. The point is fewer surprises later, not security theater now.
Best for
Security-minded builders and teams who want guardrails baked into the design.
Security Is Architecture, Not Decoration → Infrastructure and homelabs as systems-thinking laboratories
A homelab is a low-stakes place to learn the messy parts of real infrastructure — containers, networking, observability, and the failures that teach you the most. We cover what running your own systems teaches that tutorials can't. Hands-on systems thinking, not a hardware brag.
Best for
Self-taught engineers and infra-curious builders learning ops by doing.
Homelabs Teach the Messy Parts → Documentation as operational infrastructure
Docs aren't a chore bolted on at the end — they're part of how a system runs and how the next person (often future-you) recovers. We talk about writing the path down as you build, and treating runbooks and notes as load-bearing. If it isn't written down, it isn't finished.
Best for
Teams whose knowledge lives in one person's head and shouldn't.
Documentation Is Infrastructure → What gaming and community systems teach technical builders
Games and the communities around them are full of real systems thinking — incentives, feedback loops, and emergent behavior. This is a lighter, still-technical look at what running and playing in those systems teaches about building better ones. Useful for anyone who underestimates how much play sharpens engineering.
Best for
Developer, gaming, and creator communities interested in technical systems.
Gaming Taught Me Systems Thinking →